Figures
Abstract
Objective
To investigate the association of risk alleles for type 2 diabetes with prediabetes accounting for age, anthropometry, inflammatory markers and lifestyle habits.
Design
Cross-sectional study of 129 men and 157 women of medium-sized companies in northern Germany in the Delay of Impaired Glucose Tolerance by a Healthy Lifestyle Trial (DELIGHT).
Methods
Besides established risk factors, 41 single nucleotide polymorphisms (SNPs) that have previously been found to be associated with type 2 diabetes were analyzed. As a nonparametric test a random forest approach was used that allows processing of a large number of predictors. Variables with the highest impact were entered into a multivariate logistic regression model to estimate their association with prediabetes.
Results
Individuals with prediabetes were characterized by a slightly, but significantly higher number of type 2 diabetes risk alleles (42.5±4.1 vs. 41.3±4.1, p = 0.013). After adjustment for age and waist circumference 6 SNPs with the highest impact in the random forest analysis were associated with risk for prediabetes in a logistic regression model. At least 5 of these SNPs were positively related to prediabetic status (odds ratio for prediabetes 1.57 per allele (Cl 1.21–2.10, p = 0.001)).
Conclusions
This explorative analysis of data of DELIGHT demonstrates that at least 6 out of 41 genetic variants characteristic of individuals with type 2 diabetes may also be associated with prediabetes. Accumulation of these risk alleles may markedly increase the risk for prediabetes. However, prospective studies are required to corroborate these findings and to demonstrate the predictive value of these genetic variants for the risk to develop prediabetes.
Citation: Zyriax B-C, Salazar R, Hoeppner W, Vettorazzi E, Herder C, Windler E (2013) The Association of Genetic Markers for Type 2 Diabetes with Prediabetic Status - Cross-Sectional Data of a Diabetes Prevention Trial. PLoS ONE 8(9): e75807. https://doi.org/10.1371/journal.pone.0075807
Editor: Kathrin Maedler, University of Bremen, Germany
Received: April 12, 2013; Accepted: August 21, 2013; Published: September 30, 2013
Copyright: © 2013 Zyriax et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Basic funding was provided by the legal health insurence AOK NordWest, Kiel-Wellsee, Germany. Additional support for the analysis of this paper came from Merck KGaA, Darmstadt, Germany as an unrestricted research grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have the following interests. Basic funding was provided by the legal health insurence AOK NordWest, Kiel-Wellsee, Germany. Additional support for the analysis of this paper came from Merck KGaA, Darmstadt, Germany as an unrestricted research grant. Ramona Salazar and Wolfgang Hoeppner are employed by Bioglobe GmbH and Christian Herder is a PLOS ONE Editorial Board member. There are no patents, products in development or marketed products to declare. This does not alter their adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
Introduction
The prevalence of type 2 diabetes is dramatically increasing and represents a worldwide growing health problem [1]–[4]. Even in those individuals with prediabetes the risk for cardiovascular disease and total mortality is almost doubled [5]–[10]. A prediabetic status is also associated with microvascular complications [11]. Finally, 5 to 10% of untreated prediabetic patients will develop diabetes each year [12], [13]. Yet, the same proportion may convert back from the prediabetic status to normoglycemia [11]. In routine medical practice, prediabetes is not yet recognized nor treated, although it has been repeatedly demonstrated that the transition to type 2 diabetes can be delayed or avoided [14]. A sedentary lifestyle and an unhealthy dietary pattern promote weight gain, particularly central adiposity, which increases the risk for prediabetes and eventually type 2 diabetes [12], [13]. Adipocytokines such as leptin and adiponectin or the proinflammatory cytokine interleukin-6 (IL-6) are affected by lifestyle habits and seem to play an important role for weight development, body composition and risk for type 2 diabetes [15]–[20]. In addition, a family history of type 2 diabetes markedly increases the risk for diabetes reflecting the interaction of genetic factors with modern lifestyle and anthropometry [21], [22]. Results from family studies and research in various ethnic groups indicate that the heritability of the disease may exceed 50% [23]. Still, up to now information on the impact of multiple gene loci as to the risk for prediabetes is limited. In a cohort of non-diabetic Caucasians a significant association between impaired glucose tolerance and risk alleles for type 2 diabetes has been shown for female and obese individuals, whereas it has not been possible to demonstrate an effect in male, lean and insulin sensitive subjects [24].
Until 2011 approximately 40 diabetes-associated genes had been identified [23], [25]–[28]. Single nucleotide polymorphisms (SNPs) are the most commonly investigated type of specific genetic variants. However, the identification of a single gene variant associated with a complex disease such as diabetes among a large number of SNPs by statistical methods such as logistic regression analysis has limitations [29]–[31]. As more SNPs and interaction terms are added, the model becomes unstable in the sense that the variance of the parameter estimates becomes excessively large or even inestimable, when the number of model parameters exceeds the number of cases, and the effect of a genetic variant can be neutralized by the interaction with related parameters. Lately, new nonparametric predictive models have been developed to overcome this problem such as the random forest analysis, which attracts growing interest. One major advantage of this statistical approach is its capability to cope with a large number of predictors and to identify those factors with a relevant contribution to the disease, even in the presence of high order interactions [31]–[33].
This prompted us to analyze cross sectional data of young employees in the Delay of Impaired Glucose Tolerance by a Healthy Lifestyle Trial (DELIGHT) as to the association of 41 SNPs indicating risk for type 2 diabetes with a prediabetic status [23], [25]. This research uses random forest analysis to identify genetic markers of prediabetes that may add to the information of anthropometric data, inflammatory markers and lifestyle factors as to the risk for developing prediabetes [31]–[33].
Methods
Ethics Statement
The study protocol was approved by the ethical committee of Hamburg and conducted according to the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants. The trial was registered in the German Clinical Trials Register No. DRKS00000695 (www.germanctr.de).
Design and Recruitment
DELIGHT is a feasibility study on sustainable prevention of diabetes in young men and women. 18–65 year-old employees of 5 medium-sized companies in the northern part of Germany were informed about prediabetes, risk for diabetes, and chance of lifestyle modification [34]. Employees were advised how to measure their waist circumference, and were eligible for a check-up, if the waist circumference was ≥80 cm for women and ≥94 cm for men or close to these cut-off points.
Exclusion criteria were known pregnancy, known type 1 or type 2 diabetes, or acute malignant or severe chronic diseases. The final study population comprised 300 participants. However, the present analysis focuses on the data at screening of 129 men and 157 women for whom complete information about lifestyle habits, anthropometric parameters, laboratory values and genetic data were available.
Data Collection
Assessment of anthropometric data and lifestyle.
Height and weight - light clothing, but no shoes allowed - were measured to the nearest 0.5 cm or 0.1 kg, respectively, and body mass index (BMI) was calculated as BMI = (weight, kg)/(height, m)2. Waist circumference was measured in the middle between the lower rib margin and the iliac crest. Central obesity was defined by a waist circumference ≥80 cm in women and ≥94 cm in men [35].
Information on lifestyle, nutrition, socio-demographic characteristics and family history of diabetes was obtained using validated questionnaires developed for the EPIC study (European Prospective Investigation into Cancer and Nutrition), a prospective multicenter cohort study in Europe, investigating the association between lifestyle factors and chronic diseases [36]–[38]. A self-administered food questionnaire recorded the frequency and portion size of 146 food items eaten during the preceding year. Physical activity was calculated as sports in hours per week, taken into account activities during summer and wintertime. Smoking habits were described as number of cigarettes per day.
Laboratory and Clinical Data
Plasma fasting glucose and plasma glucose two hours after oral challenge with 75 g glucose (oral glucose tolerance test - OGTT) were measured from Na-fluoride-containing Monovettes (Sarstedt AG & Co, Nümbrecht, Germany). Routine laboratory parameters were determined by standard techniques in the central laboratory of the University Medical Center Hamburg-Eppendorf. Low-density lipoprotein (LDL) -cholesterol using the Friedewald formula. Prediabetes was defined as fasting blood glucose levels (IFG) between 100–<126 mg and/or plasma glucose levels two hours after an oral load of 75 g glucose (IGT) between 140–199 mg/dl. Diabetes was defined as fasting plasma glucose levels ≥126 mg/dl and/or ≥200 mg/dl two hours after 75 g of glucose [10].
Serum concentrations of IL-6 and adiponectin were measured using the Quantikine HS ELISA kit and the Quantinkine ELISA kit, respectively (R&D Systems, Wiesbaden, Germany) as described [38]. Serum leptin concentrations were determined with a bead-based assay using a Luminex 100 analyser (Luminex Corporation, Austin, TX, USA) as described [39].
Blood pressure was taken in a sitting position 3 times approximately 2 min. apart, of which the second and third value were averaged [40]. Hypertension was defined by antihypertensive medication or blood pressure ≥140 mmHg/≥90 mmHg. The homeostasis model assessment insulin resistance (HOMA-IR) score was categorized at 2.5 as the suggested upper limit of normal and ≥3.8, the upper quartile of a European population [41]–[42]. For the definition of the metabolic syndrome the criteria of the International Diabetes Federation were adopted [43].
Genetic Data
DNA was isolated from blood samples using the QIAamp DNA blood Mini Kit (Qiagen, Hilden, Germany). The gene polymorphisms of 41 identified SNPs for the risk of type 2 diabetes (Table 1) with a minor allele frequency of at least 1% in a population of European descent were analyzed by matrix assisted laserdesorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) using the Sequenom MassARRAY platform (Sequenom, San Diego, CA, USA). Assay design was performed using the standard design procedure supported by the system supplier (www.mysequenom.com). Primers were synthesized by Metabion, Martinsried, Germany and Biomers, Ulm, Germany. iPLEX GOLD application was carried out according to manufacturer’s instructions and as described previously [44]. Routinely 5% of samples were randomly picked for duplicate genotyping. The concordance was 100%.
Statistical Analyses
Baseline characteristics of the participants were reported as means and standard deviation for quantitative data and are compared between groups using Student’s t-test or Oneway ANOVA, depending on the number of groups. Qualitative scales are reported as counts and proportions and compared using chi-squared tests. Selected quantitative data like waist or BMI, HOMA-IR were also reported using discretized versions with clinically defined cut points. P-values below 0.05 were considered statistically significant. The random forests approach, a collection of classification trees, was used to cope with the large number of variables and select those markers with a relevant contribution to the defined outcome variable ‘prediabetic status’. The model of the random forests approach has been described previously in detail [32], [33]. Statistical calculations were performed running the software version R2.15.1 using the forest procedure from the party package [45]–[48]. The most promising variables found in this analysis were used to set up a multivariate logistic regression model to estimate the predictive value of identified parameters individually and collectively.
Results
Baseline Characteristics
About one third of the study population, men and women likewise, were affected by prediabetes, identified by elevated fasting and/or 2-h glucose (Table 2). Mean age and body mass index (BMI) did not differ between men and women (data not shown). However, women were characterized more often by an elevated waist circumference, and by higher levels of leptin and adiponectin, but also by a more favorable lipid profile. The prevalences of hypertension and prediabetes were comparable in both sexes. As to dietary and lifestyle habits women were characterized by a lower intake of energy, fat, saturated fat and fiber than men. Smoking habits did not differ between sexes, whereas females reported less physical activity.
Participants with prediabetes were older, had a higher HOMA-IR, and a higher BMI particularly within the category of obesity compared with normoglycemic subjects (Table 2). Also, triglyceride levels and the rate of hypertension were significantly higher in prediabetic individuals, whereas plasma concentrations of LDL- and HDL-cholesterol were similar. No differences were observed as to mean values of leptin, adiponectin, and IL-6, but also dietary intake and physical activity (data not shown).
Genetic Data
Within the study population the total number of risk alleles did not differ between men and women (Table 3). Individuals with prediabetes were characterized by a slightly, but significantly higher number of total risk alleles (Table 3). Categories of HOMA-IR did not differ in the number of risk alleles, but trends towards higher numbers of risk alleles for increasing categories of fasting glucose levels and 2-h glucose levels were observed (Table 3).
To identify genetic markers with a greater contribution as to risk for prediabetes the random forest approach was used. 41 SNPs for which associations with type 2 diabetes were published until 2011, and as well age, sex, anthropometric data, inflammatory markers (leptin, adiponectin, IL-6) and lifestyle factors known to contribute to diabetes (total energy intake, fat intake, intake of saturated fat and fiber) were included in the model.
In this explorative study those markers that exceeded the random fluctuation around zero - the magnitude of which is indicated by the negative variation and the dotted line in Figure 1 - were selected for further analyses as suggested by [49]. Markers of relevance comprised age, waist circumference, and leptin, but also 6 SNPs: rs972283 in KLF14, rs5945326 in DUSP9, rs13266634 in SLC30A8, rs10923931 in NOTCH2, rs4457053 in ZBED3, and rs1111875 in HHEX (Figure 1).
According to the analysis obtained by the random forest model the 6 SNPs representing the most powerful genetic markers were selected. Since the random forest approach does not distinguish whether the identified SNP may increase or decrease susceptibility for the disease, a logistic regression was performed including age, sex, categories of waist circumference and the 6 selected SNPs (Figure 2a). The results indicate that sex was not associated with increased risk for prediabetic status, whereas age and central obesity, particularly a waist circumference ≥88 cm in women and ≥102 cm in men, were significantly related to a higher risk. The majority of SNPs showed a tendency towards a higher risk as to prediabetic status, which was significant in carriers of rs972283 in KLF14, rs5945326 in DUSP9, and rs13266634 in SLC30A8 (Figure 2a). However, rs10923931 in NOTCH2 was significantly associated with a lower risk as to prediabetic status. To calculate the effect per risk allele, in the next step a logistic regression was performed including age, sex, categories of waist circumference and the 6 identified SNPs as sum score (Figure 2b). With every SNP the odds for prediabetes increased by 57% (Cl 1.21–2.10, p = 0.001). Evaluation as to hetero- and homozygote carriers showed similar results (data not shown). Exclusion of rs10923931 in NOTCH2 would lead to 92% (Cl 1.43–2.68, p<0.0001) increase in risk per allele. Including leptin in the analysis did not change the results (leptin 1.01 Cl 0.97–1.05; allele score 1.93 Cl 1.44–2.72).
Discussion
Compared to type 2 diabetes, the information on the impact of multiple gene loci as to the risk for prediabetes is limited and needs further clarification. Analysis of the DELIGHT data indicates that genetic variants, which predispose individuals to type 2 diabetes, may serve as risk markers for the development of prediabetes as well. Individuals with prediabetes were characterized by a significantly higher number of risk alleles than normoglycemic subjects. On average each relevant SNP increased the odds for prediabetes by 57%. Accumulation of these risk alleles may lead to a markedly increase of the risk for prediabetes, the extent of which certainly needs to be determined in adequately sized prospective studies.
There is strong evidence that a prediabetic status is sufficient to increase the risk of cardiovascular disease and death substantially [5]–[10]. Notably, in DELIGHT one third of the relatively young and healthy employees pre-selected by an elevated waist circumference was affected by prediabetes, most of them as part of a metabolic syndrome. Once identified, successful lifestyle intervention trials clearly show that diabetes may be delayed, if not prevented. Untreated, 5 to 10% prediabetic patients may develop diabetes each year [12], [13]. Therefore early detection of individuals at risk is a major challenge. Obviously, genetic markers can be determined early in life, an advantage compared to established risk factors, which confer an elevated risk primarily at a later stage. In DELIGHT the impact of well-established risk factors such as lifestyle habits or inflammatory markers appeared to be rather small, possibly explained by an elevated waist circumference as an inclusion criterion. Since age and anthropometry though risk factors for prediabetes, lack specificity, an array of simple genetic markers may be helpful to identify individuals at risk.
In the present analysis a slightly, but significantly higher number of total risk alleles characterized individuals with prediabetes compared to normoglycemic subjects. This finding was not sex-linked. In the TUEbingen Family study (TUEF) Lindner et al. reported that genetic risk alleles predict risk for impaired glucose tolerance [24]. This was only shown for women and obese individuals, yet. However, at that time the results were based on only 9 selected diabetes-associated genes, particularly those related to impaired glucose tolerance. Differences as to the influence of sexes may be explained by an underrepresentation of males in the TUEbingen Family study. This is supported by some studies which found that sex-hormones differently modulate glycemic status and IGT is more frequent in males, whereas IFG occurs more often in females [50]–[51].
Observational studies indicate that both parameters, elevated fasting and 2-h glucose values after an OGTT, seem to be strong predictors of diabetes incidence [52]–[59]. However, one should properly distinguish between variants obtained from genome-wide studies focusing on type 2 diabetes and those genes examined in epidemiological studies that are responsible for the regulation of glucose levels within the normal range [23], [60]. In DELIGHT clinically established categories of IFG and IGT showed a tendency towards a higher number of risk alleles. Yet, the risk of diabetes may be higher in subjects with isolated IGT compared to those with isolated IFG [55]. Pathophysiological mechanisms of isolated IFG and isolated IGT probably differ, but the finding and its clinical relevance need further clarification [11]. The small proportion of individuals with isolated elevated IGT in the DELIGHT project may be a matter of both the inclusion criteria of an elevated waist circumference and a low threshold of 100 mg/dl for IFG in contrast to 110 mg/dl as formerly used.
Results from the random forest analysis indicate that particularly age and waist circumference, but also leptin and 6 single-nucleotide polymorphisms out of 41 are associated with an elevated risk for prediabetes. The impact of age, waist circumference and leptin is in line with other investigations on risk for diabetes. Interestingly, sex, adiponectin, interleukin-6 levels and lifestyle habits were not selected as markers with a particularly important contribution to the disease by the random forest model. As to adiponectin and lifestyle habits, one explanation might be that age, waist circumference, leptin levels and some of the risk alleles cover much of the risk common to the preselected study population [21], [61]–[66].
In a logistic regression model the majority of the 6 selected SNPs were positively associated with prediabetic status. However, a strong significant effect was only revealed in carriers of rs972283 in KLF14, rs5945326 in DUSP9, and rs13266634 in SLC30A8, explainable either by the smaller sample size of our study or pre-selection of the participants by waist circumference. Variations at KLF14, the Krueppel like factor 14, were related to type 2 diabetes and HDL-cholesterol but also basal cell carcinoma in different populations [24], [26], [67]–[73]. The effect of KLF14 is reportedly not driven by obesity, quite unlike the known BMI- and fat mass mediated effect of FTO via insulin resistance [26], [74], [75]. Additionally, rs5945326 in DUSP9, the dual specificity protein phosphatase 9, and rs13266634 in SLC30A8, the zinc transporter, were positively related to prediabetic status. Results from other investigations indicate that DUSP9 and SLC30A8 are common susceptibility loci for type 2 diabetes across various ethnicities [23], [25], [76]–[78]. Furthermore, a positive association between prediabetic status and HHEX and ZBED was revealed. These findings are supported by others who investigated the effect of the selected SNPs as to risk for type 2 diabetes [23], [26], [60], [79].
Pre-selection criteria as to central obesity within the DELIGHT project, the exclusion of participants with known type 2 diabetes, or the sample size may have biased our result. Therefore previously identified risk variants in other gene loci such as TCF7L2 or FTO failed to show an important relationship in our analysis or were even associated with a decreased risk such as NOTCH2.
Limitations and Strength of this Study
DELIGHT has limitations that need to be addressed. First, our findings are confined to those employees who voluntarily chose to take part in the program, and were characterized by central obesity or at least a waist circumference close to the threshold. Second, the sample size was rather small and therefore associations between several identified SNPs and prediabetic status may fail to reach statistical significance. However, DELIGHT is one of the first studies, to evaluate the association between a wide array of SNPs published at the time of this analysis and risk of prediabetes above and beyond established predictors. This was possible by applying the advanced statistical method of a random forest analysis. Advantage of this explorative approach is not only the capability of coping with large numbers of predictors even in the presence of complex interactions that may have any impact.
Conclusions
This explorative analysis of DELIGHT demonstrates that at least 6 out of 41 genetic variants characteristic of individuals with type 2 diabetes may be related to prediabetic status as well. With every SNP the odds for prediabetes increased significantly beyond well-established risk factors such as age and waist circumference. In the future the identification of those markers may be useful in clinical practice to identify individuals at risk at an early stage. Certainly, more research using prospective data is required to confirm these findings, obtained by the application of the method of selected random forest analysis, to establish a clinically applicable tool.
Acknowledgments
We thank Karin Röhrig and Gabi Gornitzka (both from the German Diabetes Center) for excellent technical assistance.
Author Contributions
Conceived and designed the experiments: BCZ EW. Performed the experiments: BCZ RS WH CH EW. Analyzed the data: EV. Contributed reagents/materials/analysis tools: WH CH. Wrote the paper: BCZ EW.
References
- 1. King H, Aubert RE, Herman WH (1998) Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections. Diabetes Care 21: 1414–1431.
- 2. Lam DW, LeRoith D (2012) The worldwide diabetes epidemic. Curr Opin Endocrinol Diabetes Obes 19: 93–96.
- 3. Hu FB (2011) Globalization of diabetes: the role of diet, lifestyle, and genes. Diabetes Care 34: 1249–1257.
- 4.
International Diabetes Federation (2011) IDF Diabetes Atlas, 5th edition. Bruessels, Belgium.
- 5. Coutinho M, Gerstein HC, Wang YS (1999) The relationship between glucose and incident cardiovascular events: a metaregression analysis of published data from 20 studies of 95,783 individuals followed for 12.4 years. Diabetes Care 22: 233–240.
- 6. Port SC, Goodarzi MO, Boyle NG, Jennrich RI (2005) Blood glucose: a strong risk factor for mortality in nondiabetic patients with cardiovascular disease. Am Heart J 150: 209–214.
- 7. Qiao Q, Jousilahti P, Eriksson J (2003) Predictive properties of impaired glucose tolerance for cardiovascular risk are not explained by the development of overt diabetes during follow-up. Diabetes Care 26: 2910–2914.
- 8. Zyriax BC, Boeing H, Windler E (2005) Nutrition is a powerful independent risk factor for coronary heart disease in women - The CORA study: a population-based case-control study. Eur J Clin Nutr 59: 1201–1207.
- 9. Barr EL, Zimmet PZ, Welborn TA, Jolley D, Magliano DJ, et al. (2007) Risk of cardiovascular and all-cause mortality in individuals with diabetes mellitus, impaired fasting glucose, and impaired glucose tolerance: the Australian Diabetes, Obesity, and Lifestyle Study (AusDiab). Circulation 116: 151–157.
- 10. American Diabetes Association (2010) Diagnosis and classification of diabetes mellitus. Diabetes Care 33 Suppl 1S62–S69.
- 11. Tabák AG, Herder C, Rathmann W, Brunner EJ, Kivimäki M (2012) Prediabetes: a high-risk state for diabetes development. Lancet 379: 2279–2290.
- 12. de Vegt F, Dekker JM, Jager A, Hienkens E, Kostense PJ, et al. (2001) Relation of impaired fasting and postload glucose with incident type 2 diabetes in a Dutch population: the Hoorn Study. JAMA 285: 2109–2113.
- 13. Diabetes Prevention Program (DPP) Research Group (2002) The Diabetes Prevention Program (DPP) description of lifestyle intervention. Diabetes Care 25: 2165–2171.
- 14. Lindström J, Peltonen M, Eriksson JG, Aunola S, Hämäläinen H, et al. (2008) Determinants for the effectiveness of lifestyle intervention in the Finnish Diabetes Prevention Study. Diabetes Care 31: 857–862.
- 15. Li S, Shin HJ, Ding EL, van Dam RM (2009) Adiponectin levels and risk of type 2 diabetes: a systematic review and meta-analysis. JAMA 302: 179–188.
- 16. Qi L, Meigs JB, Liu S, Manson JE, Mantzoros C, et al. (2006) Dietary fibers and glycemic load, obesity, and plasma adiponectin levels in women with type 2 diabetes. Diabetes Care 29: 1501–1505.
- 17. Zhang Z, Lanza E, Ross AC, Albert PS, Colburn NH, et al. (2011) A high-legume low-glycemic index diet reduces fasting plasma leptin in middle-aged insulin-resistant and -sensitive men. Eur J Clin Nutr 65: 415–418.
- 18. Cardellini M, Andreozzi F, Laratta E, Marini MA, Lauro R, et al. (2007) Plasma interleukin-6 levels are increased in subjects with impaired glucose tolerance but not in those with impaired fasting glucose in a cohort of Italian Caucasians. Diabetes Metab Res Rev 23: 141–155.
- 19. Zyriax BC, Algenstaedt P, Hess UF, Schöffauer M, Bamberger C, et al. (2008) Factors contributing to the risk of cardiovascular disease reflected by plasma adiponectin: data from the coronary risk factors for atherosclerosis in women (CORA) study. Atherosclerosis. 200: 403–409.
- 20. Thorand B, Zierer A, Baumert J, Meisinger C, Herder C, et al. (2010) Associations between leptin and the leptin/adiponectin ratio and incident Type 2 diabetes in middle-aged men and women: results from the MONICA/KORA Augsburg study 1984–2002. Diabet Med 27: 1004–1011.
- 21. Ortega-Azorín C, Sorlí JV, Asensio EM, Coltell O, Martínez-González MA, et al. (2012) Associations of the FTO rs9939609 and the MC4R rs17782313 polymorphisms with type 2 diabetes are modulated by diet, being higher when adherence to the Mediterranean diet pattern is low. Cardiovasc Diabetol 11: 137.
- 22. InterAct Consortium (2013) The link between family history and risk of type 2 diabetes is not explained by anthropometric, lifestyle or genetic risk factors: the EPIC-InterAct study. Diabetologia 56: 60–69.
- 23. Herder C, Roden M (2011) Genetics of type 2 diabetes: pathophysiologic and clinical relevance. Eur J Clin Invest 41: 679–692.
- 24.
Linder K, Wagner R, Hatziagelaki E, Ketterer C, Heni M, et al.. (2012) Allele summation of diabetes risk genes predicts impaired glucose tolerance in female and obese individuals. PLoS One 7: e38224. Available: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.0000045. Accessed 27 August 2013.
- 25. McCarthy MI (2010) Genomics, type 2 diabetes, and obesity. N Engl J Med 363: 2339–2350.
- 26. Voight BF, Scott LJ, Steinthorsdottir V, Morris AP, Dina C, et al. (2010) Twelve type 2 diabetes susceptibility loci identified through large-scale association analysis. Nat. Genet 42: 579–589.
- 27. Qi L, Cornelis MC, Kraft P, Stanya KJ, Linda Kao WH, et al. (2010) Genetic variants at 2q24 are associated with susceptibility to type 2 diabetes. Hum Mol Genet 19: 2706–2715.
- 28. Yamauchi T, Hara K, Maeda S, Yasuda K, Takahashi A, et al. (2010) A genome-wide association study in the Japanese population identifies susceptibility loci for type 2 diabetes at UBE2E2 and C2CD4A-C2CD4B. Nat Genet 42: 864–868.
- 29. Culverhouse R, Suarez BK, Lin J, Reich T (2002) A perspective on epistasis: limits of models displaying no main effect. Am J Hum Genet 70: 416–471.
- 30. Moore JH, Williams SM (2005) Traversing the conceptual divided between biological and statistical epistasis: systems biology and a more modern synthesis. BioEssays 27: 637–646.
- 31.
Heidema AG, Boer JM, Nagelkerke N, Mariman EC, van der A DL, et al.. (2006) The challenge for genetic epidemiologists: how to analyze large numbers of SNPs in relation to complex diseases. BMC Genet 7: 23. Available: http://www.biomedcentral.com/1471-2156/7/23. Accessed 27 August 2013.
- 32.
Lunetta KL, Hayward LB, Segal J, Van Eerdewegh P (2004) Screening large-scale association study data: exploiting interactions using random forests. BMC Genet 5: 32. Available: http://www.biomedcentral.com/1471-2156/5/32. Accessed 27 August 2013.
- 33. Bureau A, Dupuis J, Falls K, Lunetta KL, Hayward B, et al. (2005) Identifying SNPs predictive of phenotype using random forests. Genet Epidemiol 28: 171–182.
- 34. Zyriax BC, Wolf C, Schlüter A, Khattak AH, Westenhoefer J, et al. (2012) Association of cognitive dietary restraint and disinhibition with prediabetes -cross-sectional and longitudinal data of a feasibility study in German employees. Public Health Nutr 15: 860–867.
- 35. Molarius A, Seidell JC, Sans S, Tuomilehto J, Kuulasmaa P (1999) Waist and hip circumferences, and waisthip ratio in 19 populations of the WHO MONICA Project. Int J Obes Relat Metab Disord 23: 116–125.
- 36. Bohlscheid TS, Hoting I, Boeing H, Wahrendorf J (1997) Reproducibility and relative validity of food group intake in a food frequency questionnaire developed for the German Part of the EPIC project. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 26: S59–S70.
- 37. Bohlscheid TS, Hoting I, Boeing H (1997) Wahrendorf (1997) Reproducibility and relative validity of energy and macronutrient intake of a food frequency questionnaire developed for the German Part of the EPIC project. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 26: S71–S81.
- 38. Boeing H, Wahrendorf J, Becker N (1993) EPIC-Germany - a source for studies into diet and risk of chronic diseases. Ann Nutr Metab 43: 195–204.
- 39. Schöttker B, Herder C, Rothenbacher D, Roden M, Kolb H, et al. (2013) Proinflammatory Cytokines, Adiponectin, and Increased Risk of Primary Cardiovascular Events in Diabetes Patients With or Without Renal Dysfunction: Results from the ESTHER study. Diabetes Care 36: 1703–1711.
- 40. Schulze MB, Kroke A, Saracci R, Boeing H (2002) The effect of differences in measurement procedure on the comparability of blood pressure estimates in multi-centre studies. Blood Press Monit 7: 95–104.
- 41. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, et al. (1985) Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28: 412–419.
- 42. Marques-Vidal P, Mazoyer E, Bongard V, Gourdy P, Ruidavets JB, et al. (2002) Prevalence of insulin resistance syndrome in southwestern France and its relationship with inflammatory and hemostatic markers. Diabetes Care 25: 1371–1377.
- 43. Alberti KG, Zimmet P, Shaw J (2005) IDF Epidemiology Task Force Consensus Group (2005) The metabolic syndrome-a new worldwide definition. Lancet 366: 1059–1062.
- 44. Oeth P, del Mistro G, Marnellos G, Shi T, van den Boom D (2009) Qualitative and quantitative genotyping using single base primer extension coupled with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MassARRAY). Methods Mol Biol 578: 307–343.
- 45.
R Core Team (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, Available: http://www.R-project.org/. Accessed 27 August 2013.
- 46. Hothorn T, Buehlmann P, Dudoit S, Molinaro A, Van Der Laan M (2006) Survival Ensembles. Biostatistics 7: 355–373.
- 47.
Strobl C, Boulesteix AL, Zeileis A, Hothorn T (2007) Bias in Random Forest Variable Importance Measures: Illustrations, Sources and a Solution. BMC Bioinformatics, 8(25). Available: http://www.biomedcentral.com/1471-2105/8/25. Accessed 27 August 2013.
- 48.
Strobl C, Boulesteix AL, Kneib T, Augustin T, Zeileis A (2008) Conditional Variable Importance for Random Forests. BMC Bioinformatics, 9(307). Available: http://www.biomedcentral.com/1471-2105/9/307. Accessed 27 August 2013.
- 49. Strobl C, Malley J, Tutz G (2009) An introduction to recursive partitioning: rationale, application, and characteristics of classification and regression trees, bagging, and random forests. Psychol Methods 14: 323–348.
- 50. Ding EL, Song Y, Malik VS, Liu S (2006) Sex differences of endogenous sex hormones and risk of type 2 diabetes: a systematic review and meta-analysis. JAMA 295: 1288–1299.
- 51. Regitz-Zagrosek V, Lehmkuhl E, Weickert MO (2006) Gender differences in the metabolic syndrome and their role for cardiovascular disease. Clin Res Cardiol 95: 136–147.
- 52. Abdul-Ghani MA, Lyssenko V, Tuomi T, DeFronzo RA, Groop L (2009) Fasting versus postload plasma glucose concentration and the risk for future type 2 diabetes: results from the Botnia Study. Diabetes Care 32: 281–286.
- 53. Abdul-Ghani MA, Stern MP, Lyssenko V, Tuomi T, Groop L, et al. (2010) Minimal contribution of fasting hyperglycemia to the incidence of type 2 diabetes in subjects with normal 2-h plasma glucose. Diabetes Care 33: 557–561.
- 54. Balion CM, Raina PS, Gerstein HC, Santaguida PL, Morrison KM, et al. (2007) Reproducibility of impaired glucose tolerance (IGT) and impaired fasting glucose (IFG) classification: a systematic review. Clin Chem Lab Med 45: 1180–1185.
- 55. Rathmann W, Strassburger K, Heier M, Holle R, Thorand B (2209) Incidence of Type 2 diabetes in the elderly German population and the effect of clinical and lifestyle risk factors: KORA S4/F4 cohort study. Diabet Med 26: 1212–1219.
- 56. Shaw J, Zimmet P, de Courten M, Dowse G, Chitson P (1999) Impaired fasting glucose or impaired glucose tolerance. What best predicts future diabetes in Mauritius? Diabetes Care 22: 399–402.
- 57. Janghorbani M, Amini M (2009) Comparison of fasting glucose with post-load glucose values and glycated hemoglobin for prediction of type 2 diabetes: the Isfahan diabetes prevention study. Rev Diabet Stud 6: 117–123.
- 58. Unwin N, Shaw J, Zimmet P, Alberti KGMM (2002) Impaired glucose tolerance and impaired fasting glycemia: the current status on definition and intervention. Diabet Med 19: 708–723.
- 59. Rijkelijkhuizen JM, Nijpels G, Heine RJ, Bouter LM, Stehouwer CD, et al. (2007) High risk of cardiovascular mortality in individuals with impaired fasting glucose is explained by conversion to diabetes: the Hoorn study. Diabetes Care. 30: 332–336.
- 60. Dupuis J, Langenberg C, Prokopenko I, Saxena R, Soranzo N, et al. (2010) New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet 42: 105–116.
- 61. Wannamethee SG, Lowe GD, Rumley A, Cherry L, Whincup PH, et al. (2007) Adipokines and risk of type 2 diabetes in older men. Diabetes Care 30: 1200–1205.
- 62. Raynor LA, Pankow JS, Duncan BB, Schmidt MI, Hoogeveen RC, et al. (2013) Novel Risk Factors and the Prediction of Type 2 Diabetes in the Atherosclerosis Risk in Communities (ARIC) Study. Diabetes Care 36: 70–76.
- 63.
Marques-Vidal P, Schmid R, Bochud M, Bastardot F, von Känel R, et al.. (2012) Adipocytokines, Hepatic and Inflammatory Biomarkers and Incidence of Type 2 Diabetes. The CoLaus Study. PLoS One 7: e51768. Available: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0051768. Accessed 27 August 2013.
- 64. Qi L, Liang J (2010) Interactions between genetic factors that predict diabetes and dietary factors that ultimately impact on risk of diabetes. Curr Opin Lipidol 21: 31–37.
- 65. Ruchat SM, Elks CE, Loos RJ, Vohl MC, Weisnagel SJ, et al. (2009) Evidence of interaction between type 2 diabetes susceptibility genes and dietary fat intake for adiposity and glucose homeostasis-related phenotypes. J Nutrigenet Nutrigenomics 2: 225–234.
- 66. Ruchat SM, Weisnagel JS, Rankinen T, Bouchard C, Vohl MC, et al. (2009) Interaction between HNF4A polymorphisms and physical activity in relation to type 2 diabetes-related traits: results from the Quebec Family Study. Diabetes Res Clin Pract 84: 211–218.
- 67. Small KS, Hedman AK, Grundberg E, Nica AC, Thorleifsson G, et al. (2011) Identification of an imprinted master trans regulator at the KLF14 locus related to multiple metabolic phenotypes. Nat Genet 43: 561–564.
- 68.
Chasman DI, Paré G, Mora S, Hopewell JC, Peloso G et al.. (2009) Forty-three loci associated with plasma lipoprotein size, concentration, and cholesterol content in genome-wide analysis. PLoS Genet 5: e1000730. Available: http://www.plosgenetics.org/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1000730. Accessed 27 August 2013.
- 69.
Ohshige T, Iwata M, Omori S, Tanaka Y, Hirose H et al.. (2011) Association of new loci identified in European genome-wide association studies with susceptibility to type 2 diabetes in the Japanese. PLoS One 6: e26911. Available: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0026911. Accessed 27 June 2013.
- 70. Chen G, Bentley A, Adeyemo A, Shriner D, Zhou J, et al. (2012) Genome-wide association study identifies novel loci association with fasting insulin and insulin resistance in African Americans. Hum Mol Genet 21: 4530–4536.
- 71. Zhang J, Bakheet R, Parhar RS, Huang CH, Hussain MM, et al. (2011) Regulation of fat storage and reproduction by Krüppel-like transcription factor KLF3 and fat-associated genes in Caenorhabditis elegans. J Mol Biol 411: 537–553.
- 72. Ng MC, Saxena R, Li J, Palmer ND, Dimitrov L, et al. (2013) Transferability and Fine Mapping of Type 2 Diabetes Loci in African Americans: The Candidate Gene Association Resource Plus Study. Diabetes 62: 965–976.
- 73. Stacey SN, Sulem P, Masson G, Gudjonsson SA, Thorleifsson G, et al. (2009) New common variants affecting susceptibility to basal cell carcinoma. Nat Genet 41: 909–914.
- 74.
Vimaleswaran KS, Loos RJ (2010) Progress in the genetics of common obesity and type 2 diabetes. Expert Rev Mol Med 12: e7. Available: http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=7293768. Accessed 27 August 2013.
- 75. Liu Y, Liu Z, Song Y, Zhou D, Zhang D, et al. (2010) Meta-analysis added power to identify variants in FTO associated with type 2 diabetes and obesity in the Asian population. Obesity (Silver Spring) 18: 1619–1624.
- 76.
Fukuda H, Imamura M, Tanaka Y, Iwata M, Hirose H (2012) A single nucleotide polymorphism within DUSP9 is associated with susceptibility to type 2 diabetes in a Japanese population. PLoS One 7: e46263. Available: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0046263. Accessed 27 August 2013.
- 77. Strawbridge RJ, Dupuis J, Prokopenko I, Barker A, Ahlqvist E, et al. (2011) Genome-wide association identifies nine common variants associated with fasting proinsulin levels and provides new insights into the pathophysiology of type 2 diabetes. Diabetes 60: 2624–2634.
- 78. Sladek R, Rocheleau G, Rung J, Dina C, Shen L, et al. (2007) A genome-wide association study identifies novel risk loci for type 2 diabetes. Nature 445: 881–885.
- 79. Rees SD, Hydrie MZ, Shera AS, Kumar S, O’Hare JP, et al. (2011) Replication of 13 genome-wide association (GWA)-validated risk variants for type 2 diabetes in Pakistani populations. Diabetologia 54: 1368–1374.